Post by osckhar on 18.10.13 at 12:14:45
RAMPAGE #B003
I think everybody know this card. It is the famous Rampy sent from Hank to Vikupytel in Checz Republic for a 3Dfx party. Yupe, card was sent as dead.
I guess people will ask about rare reworks on the card (front/back). Such reworks are used for memory voltage stability. It means engineers modified the card in different RAILS for getting it stable. It affects in VDDQ & VDD memory power supply. A normal Rampy works with 2.3V & 3.2V (according the values I got here from my Rampage) and 3Dfx engineers set the voltage in both RAILS 2.9V & 2.9V.
Note- datasheet memory satates that the standar value on VDDQ & VDD is 2.5V & 2.5V with a tolereance until 2.7V in both cases. Odd to check in Rampy #B003 both voltage are upper permited.
NOTES:
E-tag: #B003 and doing reference memory reworks and values.
Card uses HYUNDAI 5.5ns and not SAMSUNG 6ns.
Status of the card is dead.
It uses a Rampage chip de-capped and then re-capped after probing silicon.
[18/10/2013]
FIRST TESTS:
Card did not want to boot up. I disassemble and assemble the test socket heatsink, cleaned all the parts with isopropanol and card booted up. I was very surprised here since I thought card was in fully dead condition. Even when I saw such artifacts I was more than happy.
ARTIFACTS:
[21/10/2013]
SECOND UPGRADE
Later test voltage in different RAILS and check&study the Rampy. I used the SW from Loeschwerg- it makes me all more easy :). I wanted to see how Rampy is able to work under mode VGA 13h/12h and I started to have some clues.
My frist feeling was related on memories QFP problem. I needed to know if some memory was dead or shorts or the problem is related another issue. I checked the #B003 with my spare Rampage chip 100% funtional and same problem are present.
Again I detected some problems on RPack. I re-solder it and what I got? :)
CARD STARTS TO SHOW CORRECT IMAGE:
NO DONGLE
WITH DONGLE
Once I fixed it, I thought, the other problem must be in the opposite side. I checked, re-checked ALL rsistor pack and all are in perfect condition.
I need to continue checking the card. All would be much easy with a SW debugger tool to isolate component until find the problem. Sadly I dont own any kind of this tool. Only SW program Dominik send me. It could be a dead memory, a broken connection memory or simply a smd component...
[28/10/2013]
THIRD UPGRADE
Some important progress here.
The only way I have for finding problem is isolating components. Once I saw what 'memory bank' is giving problem I started to check component. In simple view all was perfect but the tin solder from rework on RN pack looked very bad. So, I decided unsolder and solder component again later re-check values. What I found it the SMD resistence labeled as 100 did not have tin in one side. I reemplaced it for a 10Ohm resistor and added individual smd resistor 29.5Ohm what surprise more nice when I booted up the card.
STILL ERROR PRESENT
INTERFERENCE MEMORY - VGA MODE 13h
WINDOWS LOOKS SO
Sadly each time will be more difficult to find errors and fix them. :'(
[30/10/2013]
FOURTH UPGRADE
Now the card has much better operation. I handled to take out interference / noise lowering the VDDQ to @2.3V instead of 2.9V but still errors was present. I had to upgrade the left memory bank with new ones since the errors I was getting not always are located in same place- It gave me the clue. At least 2 memory bank was in short but with my tools impossible to isolated it better and know what specific memory is in short. Once upgraded the memory. Image quality is much better than before.
BEFORE UPGRADE LEFT MEMORY BANK.
ONCE UPGRADED LEFT MEMORY BANK
WINDOWS WITH DONGLE
WINDOWS NOT DONGLE
BIOS - SOME ERRORS
HERE ERRORS (BITS SHORT) STILL PRESENT.
HERE MEMORY CLUSTER ISOLATE- NOT BITS SHORT. NOW I KNOW WHERE TO ATTACK :)
My guess why many banks are in short can be related memory power voltage configurating out of the standard value recommended for vendor. Is not normal that more than 4 banks are in short.
If today all goes well we will have another 100% functional Rampage sample. :D Just need to sleep since too much days working at night. Very tired here but is the only moment I have time free.
[30/10/2013]
FIFTH UPGRADE
I was right- once reemplaced the right memory. Again assembled again the test socket and PC started to booting up.
NOT ERRORS PRESENT WHILE BOOTING UP. I CANT EXPLAIN WHAT FEELING I GOT IT WHILE CHECKING SCREENS
HERE THE TEST WHERE NOT SHORTS BITS PRESENT.
THEN TIME TO START UP WIN98SE AND WHAT I FOUND- VERY BAD DISSAPOINTMENT... :'( I COULD NOT BELIEVE IT. AGAIN ERRORS...
AGAIN DECIDED TO RECHECK QTSOP AND WHAT SURPRISE WHEN I SAW 2 PIN IN SHORT. TAKEN OUT THE EXCESS OF SOLDER TIN... I WAS SCARED SUCH SHORT NOT DEAD THE MEMORY BUT... :D :D :D :D
NO DONGLE
WITH DONGLE
I could not believe. That sense of peace and tranquility. In the end I got it. 8)
Here where I spent a lot of time when I have some time free. Yupe, In the untidiness there is always an order. Sometimes I'm amazed that I can find needed stuff there. LoL
MY LAB
Regards,
Oscar.
I think everybody know this card. It is the famous Rampy sent from Hank to Vikupytel in Checz Republic for a 3Dfx party. Yupe, card was sent as dead.
I guess people will ask about rare reworks on the card (front/back). Such reworks are used for memory voltage stability. It means engineers modified the card in different RAILS for getting it stable. It affects in VDDQ & VDD memory power supply. A normal Rampy works with 2.3V & 3.2V (according the values I got here from my Rampage) and 3Dfx engineers set the voltage in both RAILS 2.9V & 2.9V.
Note- datasheet memory satates that the standar value on VDDQ & VDD is 2.5V & 2.5V with a tolereance until 2.7V in both cases. Odd to check in Rampy #B003 both voltage are upper permited.
NOTES:
E-tag: #B003 and doing reference memory reworks and values.
Card uses HYUNDAI 5.5ns and not SAMSUNG 6ns.
Status of the card is dead.
It uses a Rampage chip de-capped and then re-capped after probing silicon.
[18/10/2013]
FIRST TESTS:
Card did not want to boot up. I disassemble and assemble the test socket heatsink, cleaned all the parts with isopropanol and card booted up. I was very surprised here since I thought card was in fully dead condition. Even when I saw such artifacts I was more than happy.
ARTIFACTS:
[21/10/2013]
SECOND UPGRADE
Later test voltage in different RAILS and check&study the Rampy. I used the SW from Loeschwerg- it makes me all more easy :). I wanted to see how Rampy is able to work under mode VGA 13h/12h and I started to have some clues.
My frist feeling was related on memories QFP problem. I needed to know if some memory was dead or shorts or the problem is related another issue. I checked the #B003 with my spare Rampage chip 100% funtional and same problem are present.
Again I detected some problems on RPack. I re-solder it and what I got? :)
CARD STARTS TO SHOW CORRECT IMAGE:
NO DONGLE
WITH DONGLE
Once I fixed it, I thought, the other problem must be in the opposite side. I checked, re-checked ALL rsistor pack and all are in perfect condition.
I need to continue checking the card. All would be much easy with a SW debugger tool to isolate component until find the problem. Sadly I dont own any kind of this tool. Only SW program Dominik send me. It could be a dead memory, a broken connection memory or simply a smd component...
[28/10/2013]
THIRD UPGRADE
Some important progress here.
The only way I have for finding problem is isolating components. Once I saw what 'memory bank' is giving problem I started to check component. In simple view all was perfect but the tin solder from rework on RN pack looked very bad. So, I decided unsolder and solder component again later re-check values. What I found it the SMD resistence labeled as 100 did not have tin in one side. I reemplaced it for a 10Ohm resistor and added individual smd resistor 29.5Ohm what surprise more nice when I booted up the card.
STILL ERROR PRESENT
INTERFERENCE MEMORY - VGA MODE 13h
WINDOWS LOOKS SO
Sadly each time will be more difficult to find errors and fix them. :'(
[30/10/2013]
FOURTH UPGRADE
Now the card has much better operation. I handled to take out interference / noise lowering the VDDQ to @2.3V instead of 2.9V but still errors was present. I had to upgrade the left memory bank with new ones since the errors I was getting not always are located in same place- It gave me the clue. At least 2 memory bank was in short but with my tools impossible to isolated it better and know what specific memory is in short. Once upgraded the memory. Image quality is much better than before.
BEFORE UPGRADE LEFT MEMORY BANK.
ONCE UPGRADED LEFT MEMORY BANK
WINDOWS WITH DONGLE
WINDOWS NOT DONGLE
BIOS - SOME ERRORS
HERE ERRORS (BITS SHORT) STILL PRESENT.
HERE MEMORY CLUSTER ISOLATE- NOT BITS SHORT. NOW I KNOW WHERE TO ATTACK :)
My guess why many banks are in short can be related memory power voltage configurating out of the standard value recommended for vendor. Is not normal that more than 4 banks are in short.
If today all goes well we will have another 100% functional Rampage sample. :D Just need to sleep since too much days working at night. Very tired here but is the only moment I have time free.
[30/10/2013]
FIFTH UPGRADE
I was right- once reemplaced the right memory. Again assembled again the test socket and PC started to booting up.
NOT ERRORS PRESENT WHILE BOOTING UP. I CANT EXPLAIN WHAT FEELING I GOT IT WHILE CHECKING SCREENS
HERE THE TEST WHERE NOT SHORTS BITS PRESENT.
THEN TIME TO START UP WIN98SE AND WHAT I FOUND- VERY BAD DISSAPOINTMENT... :'( I COULD NOT BELIEVE IT. AGAIN ERRORS...
AGAIN DECIDED TO RECHECK QTSOP AND WHAT SURPRISE WHEN I SAW 2 PIN IN SHORT. TAKEN OUT THE EXCESS OF SOLDER TIN... I WAS SCARED SUCH SHORT NOT DEAD THE MEMORY BUT... :D :D :D :D
NO DONGLE
WITH DONGLE
I could not believe. That sense of peace and tranquility. In the end I got it. 8)
Here where I spent a lot of time when I have some time free. Yupe, In the untidiness there is always an order. Sometimes I'm amazed that I can find needed stuff there. LoL
MY LAB
Regards,
Oscar.
